«Proceedings of Gubkin University»

Investigation of accuracy characteristics and convergence of methods for predicting occurrence of defects in pipeline using artificial intelligence

UDC: 620.19:621.43
DOI: 10.33285/2073-9028-2023-4(313)-104-117

Authors:

¹ National University of Oil and Gas “Gubkin University”, Moscow, Russian Federation
² Gazprom Diagnostics, St. Petersburg, Russian Federation

Keywords: defects, in-line flaw detection, machine learning, linear regression, Random Forest

Annotation:

This paper presents the results of applying machine learning algorithms to solve the binary problem of predicting the occurrence of a stress corrosion defect on a pipe. An expanded metric was developed, which included the parameters of the pipeline, its environment, information on surveys and repairs. This was filled with data on the main gas pipeline from the Infotech system. Training was conducted on linear regression and Random Forest algorithms, and predictive models based on implemented algorithms were obtained. To assess the correctness of the forecast, data after 2021 that were not involved in training were used. It was shown that the best results were obtained by using the Random Forest algorithm, while the accuracy of forecasting reached 76 %. The convergence of the learning process was achieved at a value of less than 10 thousand epochs for both algorithms. The filtration of the results by defect depth of more than 30 % of the wall thickness allowed to raise the value of forecast accuracy to 83 %. The paper evaluates further directions for the development of research, including the use of recurrent neural networks.

Bibliography:

1. Skrynnikov S.V. Sovremennye principy i napravlenija razvitija sistemy organizacii diagnostiki, tehnicheskogo obsluzhivanija i remonta v PAO “Gazprom” // Gazovaja promyshlennost’. – 2017. – № S2 (754). – S. 4–9.
2. Leveraging Machine Learning for Pipeline Condition Assessment / L. Hon-Gfang, X. Zhao-Dong, Z. Xulei [et al.] // Journal of Pipeline Systems Engineering and Practice. – 2023 – Vol. 14, № 3. – DOI: 10.1061/JPSEA2.PSENG-1464
3. Zavyalov A.P., Zhuchkov K.N., Vasilchenko M. Process Pipeline Strength Calculation Methodology Enhancement Using Finite-Element Method // Journal of Pipeline Systems Engineering and Practice. – 2023. – Vol. 14, № 2. – DOI: 10.1061/ JPSEA2.PSENG-1401 – EDN PLAOJD.
4. Moubray J. ‘Reliability-centered Maintenance. Second Edition’. – NY: Industrial Press Inc., 1997.
5. An investigation of mitigating the safety and security risks allied with oil and gas pipeline projects / L. Kraidi, R. Shah, W. Matipa, F. Borthwick // Journal Pipeline Sci. Eng. – 2021. – № 1. – P. 349–359. – URL: https://doi.org/10.1016/j.jpse.2021.08.002
6. Kuharzh P. Prakticheskij opyt vnedrenija sistemy kontrolja celostnosti truboprovoda v kompanii MERO // Nauka i tehnologii truboprovodnogo transporta nefti i nefteproduktov. – 2012. – № 2 (6). – S. 103–105. – EDN OZDKZH.
7. Sustainable Development for Oil and Gas Infrastructure from Risk, Reliability, and Resilience Perspectives / M. Yasir, T. Afrin, Y. Huang, N. Yodo // Sustainability. – 2023. – № 15. – № 6: 4953. – URL: https://doi.org/10.3390/su15064953
8. A walk of corporate sustainability towards sustainable development: a bibliometric analysis of literature from 2005 to 2021 / A.A. Jan, F.W. Lai, J. Siddique [et al.] // Environ Sci. Pollut. Res. – 2023. – № 30. – P. 36521–36532. – URL: https://doi.org/10.1007/s11356-022-24842-4
9. Ovodkova K.V., Zhuchkov K.N., Zav’jalov A.P. K voprosu podgotovki ishodnogo massiva informacii dlja obuchenija nejronnyh setej opredeleniju parametrov defektov truboprovodov // Trudy Rossijskogo gosudarstvennogo universiteta nefti i gaza imeni I.M. Gubkina. – 2023. – № 2 (311). – S. 85–97. – DOI: 10.33285/2073-9028-2023-2(311)-85-97 – EDN TPJNNM.
10. Issledovanie osobennostej KRN magistral’nyh gazoprovodov bol’shogo diametra / V.A. Seredenok, V.L. Onackij, V.N. Tolkacheva, R.V. Aginej // Truboprovodnyj transport: teorija i praktika. – 2016. – № 5 (57). – S. 12–16. – EDN WYPNHD.
11. Koncepcija diagnostirovanija i remonta magistral’nyh gazoprovodov v regionah s vysokoj predraspolozhennost’ju k stress-korrozii / S.V. Alimov, A.B. Arabej, I.V. Rjahovskih [i dr.] // Gazovaja promyshlennost’. – 2015. – № S2 (724). – S. 10–15. – EDN UYBAQN.
12. Vasilchenko M., Zavyalov A., Zhuchkov K. Increasing the stability of a spatially distributed information system using a robust algorithm for filtering anomalous measurements // IT in industry. –2020. – Vol. 3 (8). – Р. 1–7.
13. Guljaev A.S. Vlijanie pochv na korroziju stal’nyh trub. Modelirovanie stress-korrozionnyh processov // Analitika. – 2017. – № 6 (37). – S. 74–77. – DOI: 10.22184/2227-572X.2017.37.6.74.77 – EDN ZSUBOL.
14. Zhuchkov K.N., Zavyalov A.P. Sovershenstvovanie tehnologii vnutritrubnoj diagnostiki truboprovodov s ispol’zovaniem algoritma avtomatizirovannoj obrabotki diagnosticheskih dannyh // Nauka i tehnologii truboprovodnogo transporta nefti i nefteproduktov. – 2022. – № 12 (6). – S. 540–549. – DOI: 10.28999/2541-9595-2022-12-6-540-549 – EDN WZSILO.
15. Shatrov A.V., Pashhenko D.Je. Sravnenie klassicheskih regressionnyh modelej s modeljami, postroennymi s pomoshh'ju prodvinutyh metodov mashinnogo obuchenija // Advanced Science. – 2019. – № 1 (12). – S. 24–28. – DOI: 10.25730/VSU.0536.19.004. – EDN QANBHL.
16. Tolstov A.G. Vvedenie v informatiku sistem tehnicheskoj diagnostiki / AO “Gazprom”, Informacionno-reklamnyj centr gazovoj promyshlennosti (OOO “IRC Gazprom”). – M.: IRC Gazprom, 2007. – 487 s. – EDN QNTYVL.
17. A comparison of random forest based algorithms: random credal random forest versus oblique random forest / C.J. Mantas, Ja.G. Castellano, S. Moral-García, J. Abellán // Soft Computing – A Fusion of Foundations, Methodologies and Applications. – 2019. – Vol. 23, № 21. – P. 10739–10754. – DOI: 10.1007/s00500-018-3628-5 – EDN OJRMWU.
18. Letova M.S. Realizacija regressionnyh i klassifikacionnyh zadach s pomoshh’ju metoda Random Forest // E-Scio. – 2017. – № 8 (11). – S. 15–21. – EDN ZNGRFN.
19. Improving the accuracy of estimates of the pulse sequence period using the methodology of complete sufficient statistics / K.N. Zhuchkov, M. Vasilchenko, A.D. Zagrebneva, A.P. Zavyalov // Scientific Reports. – 2022. – Vol. 12, № 1. – P. 19932. – DOI: 10.1038/s41598-022-24457-2 – EDN JCVKPJ.
20. Bulanov V.A., Fomicheva O.E. Metod derev’ev reshenij dlja zadach binarnoj i mul’tiklassovoj klassifikacii // Inzhenernaja fizika. – 2020. – № 3. – S. 19–26. – DOI: 10.25791/infizik. 03.2020.1123 – EDN PSJNSZ.